Developing a conversational AI-driven physics tutor with multimodal capabilities to foster critical thinking among G.C.E. advanced level students in Sri Lanka

Abstract

In Sri Lanka, G.C.E. A/L Physics students face various challenges, including limited access to quality educational resources and higher private tuition costs. Although many AI driven educational tools have emerged as potential solutions, most do not explicitly focus on the local A/L Physics curriculum and often provide direct answers instead of step by-step guidance, hindering conceptual understanding and critical thinking. To address this, Phy6Bot was designed and developed as a multimodal conversational AI assistant using existing MLLMs to deliver curriculum-aligned, interactive learning experiences while promoting students' critical thinking when solving Physics problems. Further, its accuracy and effectiveness were evaluated by demonstrating it to the relevant students and teachers and obtaining their feedback. Due to time and resource constraints, this study focuses solely on the Magnetic Field unit, serving as a proof-of-concept. Future iterations aim to expand across the full A/L Physics curriculum.